Point-cloud explainer¶
Under construction
PointCloudExplainer is exported by XWhy, but its current explain() method raises NotImplementedError.
Planned documentation will cover:
- accepted point-cloud formats and tensor shapes;
- point grouping or segmentation;
- removal, masking, and displacement perturbations;
- PointNet-style and custom model adapters;
- 3D attribution visualisation;
- fidelity, stability, and runtime evaluation.
Legacy point-cloud examples exist in the repository, but they should be reviewed and migrated before being treated as current package documentation.